Robotic Process Automation for Resume Processing System

Abstract

Recruiting a right candidate for a particular job is a tedious process as shortlisting few applications among thousands involves huge man power. Once the candidates are short listed then candidates have to go through different levels of interviews. Now a day's companies relay on third party for conducting the initial round of interviews. This paper aims to automate such interview process to help in recruitment with the help of robotic process automation and machine learning techniques. RPA bot monitors the mail and the file storage, if any new mail or file are received, the attachment with respect to the mails are downloaded and document classification is done on the attachments to find if they are resumes, text is extracted from the attachments, Name Entity Recognition (NER) is applied to get the important details in the attachments and the resumes are short listed based on the skills, experience and the education background. The precision value for text classification done with Naive Bayes Classifier is 92.08%, for document classification with 150k Bag of Words on BERT is 91.36%, for custom trained NER is 86.2% and finally for web scranning is 91.56%.